AI can improve efficiency, but professional scepticism — the same discipline applied to client figures — must be applied to AI outputs. The firms building a competitive advantage are the ones with governance frameworks that allow deliberate, confident, auditable AI use.
The Institute of Chartered Accountants in England and Wales released its AI guidance for members in 2024. The headline message: AI can improve efficiency, but professional scepticism — the same discipline applied to client figures — must be applied to AI outputs.
Some accounting firms have responded by banning AI entirely. Others have adopted it wholesale. Both positions carry risk. The firms building a competitive advantage are the ones that have governance frameworks allowing deliberate, confident, auditable AI use.
That is what this workshop builds.
AI bookkeeping tools have become genuinely capable. They also make confident errors: miscategorising transactions, applying wrong VAT treatments, missing reclaims. A busy practice that treats AI categorisation as final without review is building errors into client accounts systematically. The client signs the accounts. You advised on them.
Large language models can produce plausible-looking tax guidance at speed. The problem is plausibility — it looks right, it reads well, and it is sometimes wrong in ways that are not obvious to a client. When AI-drafted advice goes out on your letterhead, your professional liability follows it.
When client financial data is pasted into a general-purpose AI assistant, who processes it, where it is stored, and for how long are questions with GDPR answers. Many firms are inadvertently running client data through AI systems with no data processing agreement in place.
As MTD expands, more AI-assisted processes interact directly with HMRC systems. An error introduced by an AI tool in a MTD submission is not an IT problem — it is a professional failure. Governance frameworks need to cover the full chain from AI-assisted preparation to submission.
ICAEW AI guidance and how it maps to fundamental principles. Professional scepticism as a methodology for AI output review. Where professional liability currently sits.
We review your current AI-assisted processes — bookkeeping tools, drafting aids, practice management AI, client portals — and map specific risks to each. You leave with a risk register you can actually use.
GDPR obligations when AI processes client financial data. Vendor assessment: what a data processing addendum must contain. How to audit your current AI tool stack for compliance gaps. Client disclosure obligations.
Practical, time-efficient review workflows for AI-generated outputs. How to apply professional scepticism without losing the efficiency gains. Escalation triggers — when AI output requires enhanced human review.
Drafting your firm’s AI use policy. How to communicate AI usage to clients in a way that maintains trust. Staff training approach for different seniority levels.
| Use case | Efficiency gain | Primary risk |
|---|---|---|
| Automated bookkeeping categorisation | Significant time saving | Miscategorisation, VAT errors |
| Draft tax return preparation | Speed, completeness checks | Errors presented as final |
| Client correspondence drafting | Admin efficiency | Advice quality, tone, privilege |
| Practice management AI | Workflow, scheduling | Data handling |
| Audit evidence review | Volume handling | Missed anomalies |
| Financial forecasting tools | Analysis speed | Model assumptions, overconfidence |
We assess where you stand against your sector’s regulatory floor and identify your highest-priority governance gaps.
One hour with leadership. Sector-specific regulatory framework, immediate priority actions, the language to take this to the wider team.
Half or full day. Full risk mapping for your sector, governance protocols, AI use policy draft, and 30 days advisory access. The flagship engagement.
Multi-site, network, and group pricing available on request.
HMRC recognition means the software can interact with HMRC systems correctly. It says nothing about the accuracy of AI categorisation, the appropriateness of VAT treatments, or the data protection obligations that arise when client data passes through the platform. These are your responsibility, not HMRC’s endorsement.
A general IT security policy typically covers access controls, device management, and incident response. It was almost certainly written before the current generation of AI tools existed. AI governance requires additional layers: output review protocols, data processing obligations, professional standards compliance, and clear guidance on which tools staff can use and how. These are gaps a standard IT policy does not fill.
Be direct. Clients increasingly ask, and a clear, confident answer — “yes, we use AI tools to improve efficiency, governed by these protocols” — builds more trust than a vague non-answer. We help you draft the client-facing language.
Tell us which AI tools your firm is using and we’ll give you an honest assessment of your governance position. No sales pressure. If your governance is sound, we’ll tell you.